scTACL: 17 Benchmarking Single-Cell Transcriptomics Datasets

17 benchmarking single-cell transcriptomics datasets accompany the scTACL study, a multitask topology-aware contrastive learning framework. The data includes raw expression matrices and cell labels for evaluating tasks like data imputation and batch effect correction. Author Murong Zhou uploaded the 5.2 GB collection to figshare in May 2026 under a CC-BY-4.0 license.

Use Cases

• Benchmarking data imputation methods based on raw single-cell expression matrices
• Evaluating cell clustering algorithms using the provided cell labels
• Testing batch effect correction techniques across the multiple datasets
• Analyzing cell-cell interactions based on the topological structure of the data

Strengths

• Includes 17 distinct benchmarking datasets, providing a multi-source evaluation suite
• Contains raw expression matrices and corresponding cell labels, enabling supervised and unsupervised tasks
• Dataset size is 5.2 GB, indicating substantial data volume for model training and testing

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment for specific scale requirements

Provenance

Source

figshare, uploaded by author Murong Zhou

Collection Method

Collected as benchmarking datasets for the scTACL study, likely from public repositories

Freshness

Last updated 2026-05-29 06:42:09